Knowledge that Transforms

One of the great intellectual achievements of the twentieth century is the Bayesian theory of rational behavior under risk and uncertainty. Many economists, however, are still unaware of how strong the case really is for Bayesian theory, and many more fail to appreciate the far-reaching implications the Bayesian concept of rationality has for ethics and welfare economics. The purpose of this paper is to argue that the Bayesian rationality postulates are absolutely inescapable criteria of rationality for policy decisions; and to point out that these Bayesian rationality postulates, together with a hardly controversial Pareto optimality requirement, entail utilitarian ethics as a matter of mathematical necessity.

This paper develops an approach to understanding voluntary collective action. A simple model illustrating this approach predicts Pareto optimal provision of a nonexcludable public good in the case of identical actors with perfect information, regardless of the number of actors. In this approach, actors voluntarily subsidize each other's contributions to the provision of a public good. Each actor individually finds it optimal to match other actors' contributions dollar for dollar, and this matching behavior leads to a Pareto optimal outcome from the viewpoint of the group as a whole. The approach developed here differs from two other sets of proposed solutions to the free-rider problem. One set of proposed which may be called solutions, simply assert that individuals are forced to contribute toward the provision of collective goods, once desired quantities of such goods are known (for example, Mancur Olson; Gary Becker's theory of collusion; Theodore Groves and John Ledyard). These however, beg the question of how the coercion itself is financed, since the policing of collective agreements is itself a public good: noncontributors cannot be excluded from benefiting from the public good resulting from the coercion. Other proposed solutions of the problem of voluntary collective action assume some special property of the public good. Olson's by-product solution assumes that the public good can be jointly produced with a private good, and that the private good cannot be produced as cheaply without also producing the public good. George Stigler's asymmetry solution, as a second example, assumes that individuals have differing interests regarding the exact form that the public good will take, leading them to contribute so that the good that is provided is optimal from their own individual viewpoints. Both of these solutions implicitly introduce some form of private-ness into the public good whose provision they try to explain. This paper avoids limiting assumptions of special characteristics of public goods, and also does not postulate any coercion in the provision of the public good. After a description of the model, some examples of the predicted matching behavior and experimental evidence are briefly discussed.

The author determines which revenue schedule, when applied by economic regulation to production units, will result in an optimum response. He points out that regulations must be stable for a long-enough period to be taken seriously by a firm, although they should not be considered to be immutable, and that good regulatory strategy encourages cheap firms to produce more and expensive firms to produce less. A model framework is described for determining optimal revenue function and the various dependency factors are characterized. Two components, the traditional price signal and a penalty for departure from the quantity target, comprise the optimal reward function, which means that it is not redundant for economic planners to set both prices and production quotas. This analysis can be applied to environmental economics in terms of effluent standards.

Economists are now beginning to recognize that an understanding of layoffs is crucial for a proper analysis of unemployment. In manufacturing, about 75 percent of those who are laid off return to their original More generally, among all persons classified as losers, layoffs account for about 50 percent of all unemployment spells. Temporary layoffs are an even larger fraction of cyclical changes in the number of losers. While this group includes some seasonally unemployed, most layoffs are induced by short random or cyclical fluctuations in demand. The conventional model of search unemployment is inappropriate for those and the modern theory of the Phillips curve requires substantial modification because of the size and cyclical variation of unemployment. I In a previous paper (1976), I showed analytically that our current system of unemployment insurance (UI) provides a substantial incentive for increased unemployment.2 The present paper provides micro-economic evidence that UI actually such a powerful effect. The estimates imply that the incentive provided by the current average level of UI benefits is responsible for approximately one-half of unemployment. It is important to note that the current study shows that UI increases the amount of unemployment, but does not deal with the mean per spell. This distinction deserves emphasis because nearly all previous empirical work focused the potential effect of UI duration. This focus is both unfortunate and surprising since UI can actually increase total unemployment while decreasing the mean per spell. While UI increases the of any given spell of unemployment, it may also induce more very short spells of unemployment. This possibility of reduced mean is clear in my 1976 theoretical analysis. An additional practical *Professor of economics, Harvard University. I am grateful to the National Science Foundation for support of this research, to David Ellwood and Joseph Kahan for assistance with the statistical calculations, and to Richard Freeman, Zvi Griliches, Daniel Hamermesh, James Medoff, Melvin Reder, and Jeffrey Sachs for discussions and comments. Earlier versions of this paper were presented at seminars at Chicago, Harvard, and Yale universities. IIn my 1975 paper, pp. 737-42, I discuss the implications of layoffs for the theory of search unemployment, the Phillips curve, and wage inflexibility. Although the standard criterion of unemployment is active seeking within the past four weeks, individuals are officially classified as unemployed without any inquiry about recent job-seeking activity if they state that they are on awaiting recall by their employers. Some of those look for jobs or alternative permanent employment, but the vast majority do return to their original Readers should not be confused by the two quite separate meanings of the term layoff in the Department of Labor's lexicon. In manufacturing establishment data, a is a separation initiated by the employer (not a quit) and may be permanent or In the Current Population Survey (CPS), an individual is if he is not working but has a job to which he is expecting to be recalled by his employer. To emphasize that I am dealing with those layoffs expected to terminate in recall, I use the adjective temporary. Unfortunately, the CPS uses the word in a different and quite confusing way: persons are divided into an indefinite duration group (in which the individual does not have an expected date of recall within thirty days) and a temporary group (when such a date is known). When it is useful to distinguish these groups, I use the terms indefinite duration and fixed duration; in my usage, the term includes both groups. 2My 1976 paper is really an explicit proof of arguments made more informally in my earlier study for the Joint Economic Committee (1973). For a similar development, see Martin Baily.